A mechanistic forward osmosis model based on nonideal principles and the continuity equation was adapted to the dry salting of salmon. The novelty of this model is that the water loss is coupled to the salt uptake by means of the activity gradient. Consequently, besides the primarily desired predictive purposes, the model also explains why the ion uptake triggers the osmotic dehydration. The determination of the model parameters, as well as the validation of the model was carried out by comparing the results of the simulations with experimental salt and water concentration distributions. The good predictions of the model allow the establishment of a tool to have a better control of the time the salting process must last to meet both organoleptic and safety requirements. Additionally, it is transversally applicable to other food matrices, and by extension, to other engineering situations involving dehydration induced by ion uptake. Practical applications An increasing salt concentration in muscle tissue affects two very important aspects: the water loss (yield) and the water activity (consumer safety). The model presented in this work describes how these three variables are related by means of a physics‐based link. This allows its use to optimize the process. Moreover, as this model can also predict the water activity distribution at any time, it also helps to ensure that every point in the system meets the safety requirements.